Re: + kthread-numa-aware-kthread_create_on_cpu.patch added to -mm tree

[Andrew Morton <akpm@linux-foundation.org>]

On Wed, 19 Jan 2011 21:50:47 +0100
Eric Dumazet <eric.dumazet@gmail.com> wrote:

> Le mercredi 19 janvier 2011 __ 12:44 -0800, Andrew Morton a __crit :
> > On Wed, 19 Jan 2011 21:21:55 +0100
> > Eric Dumazet <eric.dumazet@gmail.com> wrote:
> > 
> > > Le mercredi 19 janvier 2011 __ 12:17 -0800, Andi Kleen a __crit :
> > > > > 
> > > > > I wont discuss about a function _name_ Andrew.
> > > > > 
> > > > > If nobody can chose one better than the one I chosed, what can I do ?
> > > > 
> > > > kthread_create_node() would seem logical (with a node parameter)
> > > > 
> > > > -Andi
> > > 
> > > I already explained why I dont like this suggestion.
> > > 
> > > 1) My plan was to later add cpu affinity.
> > 
> > That would be called kthread_create_on_cpu() except whoops, this patch
> > already took that.
> > 
> 
> I surrender :)

Does that mean we have a name ;)

> I'll send a patch, or do you prefer I respin the 4 patches ?

I can trivially edit the patches locally if it's just a rename. 
kthread_create_for_cpu() would do the trick, I suggest.

If we decide on kthread_create_node(node_t) then that's a significant
rework.

I'm all worn out too and would be OK with either approach.